We're gotta work with whatever the hospitals give us.
The most common codes we see are CPT/HCPCS codes (coupled with revenue codes), or DRG codes. When the hospital provides any one of REV, CPT, HCPCS, MS-DRG, APR-DRG, APC, NDC, or EAPG, we break them out into their own columns.
We keep all the code metadata -- that is, we leave the code strings mostly intact, and then extract the code itself into a separate column. This preserves things like "CPT4" or whatever, as a string, but we do not have a way yet of splitting CPT codes into multiple subtypes. So we haven't lost any data, but we haven't totally figured out how to query it perfectly well either.
